Demonstration of the Excitation and Control of Plasma ... · Demonstration of the Excitation and Control of Plasma Wakeﬁelds by Multiple Laser Pulses Simon Hooker, C. Arran, L

Demonstration of the Excitation and Control of Plasma Wakefields by Multiple Laser Pulses

Simon Hooker, C. Arran, L. Corner, J. Cowley, G. Cheung, J. Holloway, R.J. Shalloo, C. Thornton, R. Walczak John Adams Institute for Accelerator Science & Department of Physics, University of Oxford

C. Gregory and D.R. Symes Rutherford Appleton Laboratory

N.H. Matlis Centre for Free-Electron Laser Science, DESY, Hamburg

S.P.D. Mangles John Adams Institute for Accelerator Science & Imperial College, London

Demonstration of the Excitation and Control of Plasma Wakefields by Multiple Laser Pulses

Simon Hooker, C. Arran, L. Corner, J. Cowley, G. Cheung, J. Holloway, R.J. Shalloo, C. Thornton, R. Walczak John Adams Institute for Accelerator Science & Department of Physics, University of Oxford

C. Gregory and D.R. Symes Rutherford Appleton Laboratory

N.H. Matlis Centre for Free-Electron Laser Science, DESY, Hamburg

S.P.D. Mangles John Adams Institute for Accelerator Science & Imperial College, London

Excitation and Control of Plasma Wakefields by Multiple Laser Pulses

J. Cowley,1 C. Thornton,1 C. Arran,1 R. J. Shalloo,1 L. Corner,1 G. Cheung,1 C. D. Gregory,2

S. P. D. Mangles,3 N. H. Matlis,4 D. R. Symes,2 R. Walczak,1 and S. M. Hooker1,*1John Adams Institute for Accelerator Science, University of Oxford,

Denys Wilkinson Building, Keble Road, Oxford OX1 3RH, United Kingdom2Central Laser Facility, Rutherford Appleton Laboratory, Didcot OX11 0QX, United Kingdom3John Adams Institute for Accelerator Science, Blackett Laboratory, Imperial College London,

London SW7 2AZ, United Kingdom4Deutsches Elektronen-Synchrotron (DESY), Notkestraße 85, Hamburg 22607, Germany

(Received 15 March 2017; revised manuscript received 31 May 2017; published 27 July 2017)

We demonstrate experimentally the resonant excitation of plasma waves by trains of laser pulses. Wealso take an important first step to achieving an energy recovery plasma accelerator by showing that aplasma wave can be damped by an out-of-resonance trailing laser pulse. The measured laser wakefields arefound to be in excellent agreement with analytical and numerical models of wakefield excitation in thelinear regime. Our results indicate a promising direction for achieving highly controlled, GeV-scale laser-plasma accelerators operating at multikilohertz repetition rates.

DOI: 10.1103/PhysRevLett.119.044802

Particle accelerators lie at the heart of many areas ofscience, technology, and medicine either through directapplication of the particle beams or by driving radiationsources such as synchrotrons and free-electron lasers(FELs). With conventional radio-frequency (RF) techno-logy the electric field used to accelerate particles istypically less than 100 MVm−1, which is a significantfactor determining the size and cost of the machine. Indistinct contrast, plasma accelerators can generate gradientsof order 100 GVm−1, which shrinks the length of theacceleration stage by orders of magnitude.In a plasma accelerator the acceleration field is generated

within a trailing plasma wakefield excited by displacementof the plasma electrons by a driving laser pulse [1–4] orparticle bunch [5,6]. Laser-driven plasma accelerators havemade impressive progress [7] in recent years. They can nowgenerate electron beams with energies comparable to thoseused in synchrotrons and FELs (a few GeV), but inaccelerator stages only a few centimetres long [8–10],with bunch durations in the femtosecond range [11–13],and with properties ideal for generating femtosecondduration visible to x-ray pulses [14–20].In almost all recent work the plasma wakefield has been

driven by single laser pulses from high-power Ti:sapphirechirped-pulse-amplification laser systems. Unfortunately,these have very low wall-plug efficiency (<0.1%) andcannot readily operate at pulse repetition frequenciesmuch above frep ¼ 10 Hz. At present, therefore, the driverparameters severely restrict the number of potential appli-cations of laser-plasma accelerators.We recently reexamined [21] multipulse laser wakefield

acceleration (MP-LWFA) in which the wakefield is excitedby a train of low-energy laser pulses, rather than by a single,

high-energy pulse. If the pulses are spaced by the plasmawavelength λp0 ¼ 2πc=ωp0, then the wakefields driven bythe pulses in the train add coherently, causing the plasmawave amplitude to grow towards the back of the train. Here,the plasma frequency is ωp0 ¼ 2π=Tp0 ¼ ðne0e2=meϵ0Þ1=2,where ne0 is the ambient electron density. We note thatBenedetti et al. [22] have studied an alternative scheme inwhich the wakefield is driven by an incoherent combinationof laser pulses arranged longitudinally, or transversely,within a single plasma period.Using a train of low-energy laser pulses opens plasma

accelerators to novel laser technologies, such as fiber orthin-disk lasers, which cannot directly deliver joule-levelshort pulses, but which can provide lower-energy pulseswith frep in the kilohertz range, whilst achieving wall-plugefficiencies at least 2 orders of magnitude higher thanconventional solid-state lasers [23]. Our recent analysis[21] showed that a MP-LWFA driven by a near-term lasersystem of this type could drive wakefields with an accel-erating field of 4.7 GVm−1, with a dephasing-limitedenergy gain of 0.75 GeV, and that with frep ¼ 10 kHzthese could drive compact coherent and incoherent x-raysources with average brightnesses exceeding those availablefrom large scale, nonsuperconducting, RF accelerators. Afurther advantage of MP-LWFA is that it provides a naturalarchitecture for “energy recovery”: the use of one or moretrailing laser pulses to remove (and potentially recycle)energy remaining in the wakefield after particle acceleration.Energy recovery is likely to be an important capability infuture plasma accelerators operating at high average powers.In this Letter we present the first demonstration of

wakefield excitation by a laser pulse structure that is longcompared to the plasma period, and for which there is

PRL 119, 044802 (2017) P HY S I CA L R EV I EW LE T T ER Sweek ending28 JULY 2017

0031-9007=17=119(4)=044802(6) 044802-1 © 2017 American Physical Society

Simon Hooker University of Oxford

EAAC, Elba, 24 - 30 Sep 2017

Multi-pulse laser wakefield acceleration

‣ Drive wakefield with train of low-energy laser pulses

‣ Resonant excitation if pulse spacing matched to plasma period

‣ Not a new idea

• Many theory papers published in 1990s

• Similar efforts underway for PWFAs

‣ Allows use of different laser technologies

• Fibre lasers: 5.7 mJ, 200 fs @ 40 kHz [Klenke et al. Opt. Lett. 39 6875 (2014)

• Thin-disk Nd:YAG: 0.2 - 1 J, 1 ps at 5 kHz commercially available

‣ Could be route to multi-kHz repetition rates with high wall-plug efficiency

‣ Potential for additional control over wake excitation

‣ Natural architecture for “energy recovery”

S.M. Hooker et al. J. Phys. B 47 234003 (2013)

pulse train

growing plasma wave

identical electric fields

Multi-pulse LWFA Only 4 laser pulses shown. In reality would use 10 - 100!


EAAC, Elba, 24 - 30 Sep 2017

Isn’t that just beat-wave?

pulse train

growing plasma wave



EAAC, Elba, 24 - 30 Sep 2017


Yes

‣ Takes advantage of resonant excitation

‣ One way to generate train would be to interfere two, long pulses s.t. ω₂ - ω₁ = ωp

pulse train

growing plasma wave



EAAC, Elba, 24 - 30 Sep 2017


Yes



No

‣ Much easier to find and lock to resonance

• Total duration of pulse train much shorter: 1 - 30ps (i.e. N = 10 - 100), compared to 100ps - 1 ns in earlier work..

• Relative error in density ~ 1/N

• Repetition rate of lasers at least 1000× higher

‣ Not limited by relativistic detuning (Rosenbluth-Liu)

• Pulse / modulation spacing does not have to be fixed

• Deutsch et al. Phys. Fluids B 3 1773 - 1780 (1991)

pulse train

growing plasma wave



EAAC, Elba, 24 - 30 Sep 2017


Yes



No

‣ Much easier to find and lock to resonance

• Total duration of pulse train much shorter: 1 - 30ps (i.e. N = 10 - 100), compared to 100ps - 1 ns in earlier work..

• Relative error in density ~ 1/N

• Repetition rate of lasers at least 1000× higher

‣ Not limited by relativistic detuning (Rosenbluth-Liu)

• Pulse / modulation spacing does not have to be fixed

• Deutsch et al. Phys. Fluids B 3 1773 - 1780 (1991)

pulse train

growing plasma wave


See Roman Walczak’s talk WG5

Tue 18:40

gas cell

stretcherblock

400 nmspectrometer

diagnosticpulse Michelson

timingstage

pulse trainMichelson

vacuumcompressor

vacuumchamber


EAAC, Elba, 24 - 30 Sep 2017

Proof-of-principle demonstration

‣ Expts with Astra TA2 laser at RAL

‣ Astra delivers single 500 mJ, 40 fs Ti:sapphire pulses

‣ Converted single pulses into train of N = 1 - 7 pulses

‣ Wakefield measured by frequency-domain holography & TESS

Gas cell target

J. Cowley et al. Phys. Rev. Lett. 119 044802 (2017)

EL: 160 - 270 mJ w0: (35 ± 5) µm Lcell: 3 mm


EAAC, Elba, 24 - 30 Sep 2017

Frequency-domain holography

‣ Probe accumulates phase with propagationN.H. Matlis et al. Nat. Phys 2 749 (2006)

‣ Frequency maps to position ζ relative to pulse(s)

‣ Spectral interference with (identical) reference pulse allows phase shift to be determined

FDH signal

‣ Knowing the probe & ref spectra, and Δt, allows Δφ to be extracted

��p

⇡!2

p0

2!probe

c

Zne

(⇣, z)

ne0

dz

Δt

pump pulse

ref. pulseprobe pulse

plasmawakefield

�� ˜E(!)��2

=

�� ˜Eprobe

(!)��2

+

�� ˜Eref

(!)��2

+ 2

�� ˜Eprobe

(!)�� ˜E

ref

(!)�� cos (!�t+��)


EAAC, Elba, 24 - 30 Sep 2017

TESS

‣ Temporally-encoded spectral shifting (TESS)

‣ Assume plasma wave is sinusoidal

‣ Fourier transform of interferogram:

• Sideband at probe-ref pulse spacing: t = T

• Satellites at t = T ± m ωp × ψ(2) give plasma period

• Ratio of satellite(s) to sideband gives wake amplitude

��p

⇡!2

p0

2!probe

`

c

�ne

ne0

sin (!p0

⇣ + ✓)

‣ Probe contains terms of form,

exp (i��p) = exp [i� sin (!p0⇣ + ✓)]

=

1X

m=�1Jm (�) exp [im (!p0⇣ + ✓)]

N. H. Matlis et al. Opt. Lett. 41 5503 (2016)

y -p

ositi

ony

-pos

ition

Frequency

Time

spectral fringes

Fouriertransform

sideband @ pulse sep.

0 Δt


EAAC, Elba, 24 - 30 Sep 2017

TESS







��p

⇡!2

p0

2!probe

`

c

�ne

ne0

sin (!p0

⇣ + ✓)


exp (i��p) = exp [i� sin (!p0⇣ + ✓)]

=

1X

m=�1Jm (�) exp [im (!p0⇣ + ✓)]


y -p

ositi

ony

-pos

ition

Frequency

Time

spectral fringes

Fouriertransform


0 Δt

y -p

ositi

ony

-pos

ition

Frequency

Time

spectral fringes

Fouriertransform


0 Δt

ψ(2) ωp

satellites


EAAC, Elba, 24 - 30 Sep 2017

TESS







��p

⇡!2

p0

2!probe

`

c

�ne

ne0

sin (!p0

⇣ + ✓)


exp (i��p) = exp [i� sin (!p0⇣ + ✓)]

=

1X

m=�1Jm (�) exp [im (!p0⇣ + ✓)]


See Chris Arran’s poster on

Wednesday

y -p

ositi

ony

-pos

ition

Frequency

Time

spectral fringes

Fouriertransform


0 Δt

y -p

ositi

ony

-pos

ition

Frequency

Time

spectral fringes

Fouriertransform


0 Δt

ψ(2) ωp

satellites


EAAC, Elba, 24 - 30 Sep 2017

TESS with real data

Wavelength / nm375 380 385 390 395 400 405 410 415

Posi

tion

/ µm

100

200

300

Raw spectrum (no gas)


EAAC, Elba, 24 - 30 Sep 2017

TESS with real data

Wavelength / nm375 380 385 390 395 400 405 410 415

Posi

tion

/ µm

100

200

300


0 1 2 3 4 5 6 7t (ps)

-100

0

100

r (µ

m)

Fourier transform (no gas)


EAAC, Elba, 24 - 30 Sep 2017

TESS with real data

Wavelength / nm375 380 385 390 395 400 405 410 415

Posi

tion

/ µm

100

200

300


0 1 2 3 4 5 6 7t (ps)

-100

0

100

r (µ

m)

Fourier transform (with gas)


EAAC, Elba, 24 - 30 Sep 2017

TESS with real data

Wavelength / nm375 380 385 390 395 400 405 410 415

Posi

tion

/ µm

100

200

300


0 1 2 3 4 5 6 7t (ps)

-100

0

100

r (µ

m)

Fourier transform (with gas)


EAAC, Elba, 24 - 30 Sep 2017

TESS waterfall

0 1 2 3 4 5 6 7t (ps)

0

10

20

30

40

Pres

sure

(mba

r)J. Cowley et al. Phys. Rev. Lett. 119 044802 (2017)


EAAC, Elba, 24 - 30 Sep 2017

TESS waterfall

0 1 2 3 4 5 6 7t (ps)

0

10

20

30

40

Pres

sure

(mba

r)

0 1 2 3 4 5 6 7t (ps)

0

10

20

30

40

Pres

sure

(mba

r)

ωpψ(2)



EAAC, Elba, 24 - 30 Sep 2017

Single drive pulse: FDH retrieval of wakefield

‣ Wakefield clearly observed! ζ (ps)

050

-50-2

0

2

2.0 1.8 1.6 1.4 1.2 1.0

r (μ

m)

δne / ne0(%)



EAAC, Elba, 24 - 30 Sep 2017


‣ Wakefield clearly observed!


‣ Measured plasma period agrees with expected value

• P = 31 mbar ⇒ Tp = 90 fs

ζ (ps)

050

-50-2

0

2

2.0 1.8 1.6 1.4 1.2 1.0

r (μ

m)

δne / ne0(%)

10 Tp = (900 ± 50) fs


EAAC, Elba, 24 - 30 Sep 2017


‣ Wakefield clearly observed!


0 10 20 30 40

100

150

200 TESSFDHTheory

T p (fs)

Pressure (mbar)

Tp =2⇡

!p/ 1p

P

‣ Measured plasma period agrees with expected value

• P = 31 mbar ⇒ Tp = 90 fs

‣ FDH and TESS measurements consistent & agree with theory (no free parameters)

ζ (ps)

050

-50-2

0

2

2.0 1.8 1.6 1.4 1.2 1.0

r (μ

m)

δne / ne0(%)

10 Tp = (900 ± 50) fs

Wakefields excited by trains of pulses


EAAC, Elba, 24 - 30 Sep 2017

Pulse train characterization

‣ Pulse trains characterized by comparing measured single-shot autocorrelation …

‣ with SSA calculated from

• Measured laser spectrum

• Compressor geometry

R. J. Shalloo et al. Nucl. Instr. Meth. A 829 383 (2016)

Autocorrelation

Deduced pulse trainRaw SSA


EAAC, Elba, 24 - 30 Sep 2017

Two-pulse driverJ. Cowley et al. Phys. Rev. Lett. 119 044802 (2017)

-500 0 500Time (fs)

0

1

Nor

m. I

nten

sity

N = 2

�ne

ne0=

�ne

ne0

�

N=1

⇥

��sin

�12N!p0�⌧

�

sin�12!p0�⌧

��


EAAC, Elba, 24 - 30 Sep 2017


-500 0 500Time (fs)

0

1

Nor

m. I

nten

sity

10 20 30 40Pressure (mbar)

0

0.5

1

1.5

2

δ n e /

n e0 (%

)

N = 2

�ne

ne0=

�ne

ne0

�

N=1

⇥

��sin

�12N!p0�⌧

�

sin�12!p0�⌧

��


EAAC, Elba, 24 - 30 Sep 2017


-500 0 500Time (fs)

0

1

Nor

m. I

nten

sity


0

0.5

1

1.5

2

δ n e /

n e0 (%

)

N = 2

�ne

ne0=

�ne

ne0

�

N=1

⇥

��sin

�12N!p0�⌧

�

sin�12!p0�⌧

��

‣ Excellent fit to analytic expression for N = 2

• δτ = (407 ± 6) fs

• SSA: δτ = (365 ± 40) fs

• Spectra interference δτ = (420 ± 20) fs

‣ Even better agreement with fit of wake calculated from measured pulse train with ζ → αζ

‣ Find α = 1.1 ± 0.02

‣ Energy recovery!

• Wake amplitude reduced by (44 ± 8)%


EAAC, Elba, 24 - 30 Sep 2017

Multi-pulse driverJ. Cowley et al. Phys. Rev. Lett. 119 044802 (2017)

-500 0 500Time (fs)

0

1

Nor

m. I

nten

sity

N ≈ 7

�ne

ne0=

�ne

ne0

�

N=1

⇥

��sin

�12N!p0�⌧

�

sin�12!p0�⌧

��


EAAC, Elba, 24 - 30 Sep 2017


-500 0 500Time (fs)

0

1

Nor

m. I

nten

sity

N ≈ 7


0

0.5

1

1.5

2

δ n e /

n e0 (%

)

�ne

ne0=

�ne

ne0

�

N=1

⇥

��sin

�12N!p0�⌧

�

sin�12!p0�⌧

��


EAAC, Elba, 24 - 30 Sep 2017


-500 0 500Time (fs)

0

1

Nor

m. I

nten

sity

N ≈ 7

�ne

ne0=

�ne

ne0

�

N=1

⇥

��sin

�12N!p0�⌧

�

sin�12!p0�⌧

��

‣ Excellent fit to analytic expression for N = 7

• δτ = (116 ± 2) fs,

• SSA: δτ = (112 ± 6) fs

‣ Excellent agreement with fit of wake calculated from measured pulse train with ζ → αζ

‣ Find α = 1.04 ± 0.02


0

0.5

1

1.5

2

δ n e /

n e0 (%

)


EAAC, Elba, 24 - 30 Sep 2017

Summary

‣ First demonstration of MP-LWFA OR first demonstration of beat-wave with chirped driving pulses

‣ First step to energy recovery

‣ Demonstrated TESS measurements consistent with FDH & measured wake amplitudes as low as 1%

‣ Future

• Adjust pulse chirp to maintain resonance at large wake amplitudes

• Accelerate electrons!

This work was supported by: • Science and Technology Facilities Council (Grant Nos ST/J002011/1, ST/P002048/1, and 1507620) • Engineering and Physical Sciences Research Council (Grant Nos 1378575 and 1093881) • Helmholtz Association of German Research Centres (Grant No. VH-VI-503) • Air Force Office of Scientific Research, Air Force Material Command, USAF [Grant No. FA8655-13-1-2141]

0 1 2 3 4 5 6 7t (ps)

0

10

20

30

40

Pres

sure

(mba

r)

ωpψ(2)

ζ (ps)

050

-50-2

0

2

2.0 1.8 1.6 1.4 1.2 1.0

r (μ

m)

δne / ne0(%)

Documents

Demonstration of the Excitation and Control of Plasma ... · Demonstration of the Excitation and Control of Plasma Wakeﬁelds by Multiple Laser Pulses Simon Hooker, C. Arran, L